JP2021194690A - Plate thickness control device, plate thickness control method, program and rolling system - Google Patents

Abstract

To provide a plate thickness control device that can improve accuracy in predicting a rolling load.SOLUTION: A plate thickness control device, which is a plate thickness control device of a rolling machine that performs rolling of a plurality of passes, comprises: a calculating part that calculates a predicted value of a rolling load in a target pass; an obtaining part that obtains rolling-related data concerning the target pass or a pass before the target pass; a predicting part that predicts an error index from the obtained rolling-related data, in accordance with a prediction model that is created with the error index representing an error between an actual value of rolling load in rolling in the past and the predicted value set as a target variable and the rolling-related data set as an explanatory variable; a correcting part that corrects the calculated predicted value of the rolling load, in accordance with the predicted error index; and a setting part that calculates a set value of a roll gap of the rolling machine, on the basis of the corrected predicted value of rolling load.SELECTED DRAWING: Figure 1

Description

The present invention relates to a plate thickness control device, a plate thickness control method, a program, and a rolling system.

In Patent Document 1, when calculating the rolling load in each pass in multi-pass rolling, it is derived in correspondence with the calculation formula of the deformation resistance using the dislocation density as a parameter and the strain hardening and dynamic recovery during machining in each pass. Deformation resistance is calculated using the formula for calculating the dislocation density and the formula for calculating the dislocation density residual rate derived based on the formula that approximates the change in dislocation density between passes with a quadratic polynomial for the dislocation density immediately after the previous pass. A method of calculating and calculating a rolling load from this deformation resistance is disclosed.

Japanese Unexamined Patent Publication No. 8-243619

In the above method, since the deformation resistance is predicted based on the time evolution formula of the dislocation density, the calculation error of the dislocation density accumulates as the number of passes increases, and the prediction error of the deformation resistance may increase.

The present invention has been made in view of the above problems, and a main object thereof is a plate thickness control device, a plate thickness control method, a program, and a rolling system capable of improving the prediction accuracy of a rolling load. To provide.

In order to solve the above problems, the plate thickness control device according to one aspect of the present invention is a plate thickness control device for a rolling mill that rolls a plurality of passes, and is a calculation unit that calculates a predicted value of a rolling load of a target pass. The objective variable is an error index showing the error between the actual value and the predicted value of the rolling load in the past rolling, and the explanatory variable is the rolling-related data. A prediction unit that predicts an error index from the acquired rolling-related data according to the prediction model generated as, and a correction unit that corrects the predicted value of the calculated rolling load according to the predicted error index. A setting unit for calculating a set value of a roll gap of the rolling mill based on the predicted value of the corrected rolling load is provided.

Further, in the plate thickness control method of another aspect of the present invention, the predicted value of the rolling load of the target pass in the rolling of a plurality of passes by the rolling mill is calculated, and the rolling-related data regarding the target pass or the previous pass is obtained. The error index is obtained from the acquired rolling-related data according to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable and the rolling-related data as the explanatory variable. Prediction is made, the predicted value of the calculated rolling load is corrected according to the predicted error index, and the set value of the roll gap of the rolling mill is calculated based on the predicted value of the corrected rolling load.

Further, the program of another aspect of the present invention is to calculate a predicted value of the rolling load of a target pass for rolling of a plurality of passes, to acquire rolling-related data regarding the target pass or a previous pass, and to obtain rolling-related data in the past. Predicting the error index from the acquired rolling-related data according to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in rolling as the objective variable and the rolling-related data as the explanatory variable. Correcting the calculated predicted value of the rolling load according to the predicted error index, and calculating the set value of the roll gap of the rolling mill based on the corrected predicted value of the rolling load. Let the computer do it.

Further, the rolling system of another aspect of the present invention includes a rolling mill that rolls a plurality of passes, a calculation unit that calculates a predicted value of the rolling load of the target pass, and rolling-related matters relating to the target pass or a path before the target pass. According to the acquisition unit that acquires the data, the error index showing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable, and the rolling-related data as the explanatory variable, the acquired rolling-related A prediction unit that predicts an error index from data, a correction unit that corrects the calculated predicted value of rolling load according to the predicted error index, and the rolling based on the corrected predicted value of rolling load. A setting unit for calculating a set value of the roll gap of the machine and a control unit for controlling the roll gap of the rolling mill based on the calculated set value are provided.

According to the present invention, it is possible to improve the prediction accuracy of the rolling load.

It is a figure which shows the structural example of a rolling system. It is a figure which shows the procedure example of the plate thickness control method. It is a figure which shows the example of the training data.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a configuration example of the rolling system 100 according to the embodiment. The rolling system 100 includes a rolling mill 1 and a plate thickness control device 2 according to an embodiment.

The rolling mill 1 includes a pair of work rolls 11 and 12 for rolling. Specifically, the rolling mill 1 is a reverse type rolling mill that reciprocates a thick steel plate SP between a pair of work rolls 11 and 12 to perform rolling of a plurality of passes (so-called multi-pass rolling).

The plate thickness control device 2 drives a rolling device (not shown) provided in the rolling mill 1 to adjust the roll gap between the pair of work rolls 11 and 12.

The plate thickness control device 2 is a computer including a CPU, RAM, ROM, non-volatile memory, an input / output interface, and the like. The CPU of the plate thickness control device 2 executes information processing according to a program loaded from the ROM or the non-volatile memory into the RAM.

The program may be supplied via an information storage medium such as an optical disk or a memory card, or may be supplied via a communication network such as the Internet or a LAN.

The plate thickness control device 2 includes a load prediction value calculation unit 21, a request point data acquisition unit 22, a storage data acquisition unit 23, a softening correction coefficient calculation unit 24, a roll gap set value calculation unit 25, and a control unit 26. .. These are realized by the plate thickness control device 2 executing information processing according to a program.

The load prediction value calculation unit 21 is an example of the calculation means, the request point data acquisition unit 22 is an example of the acquisition means, and the softening correction coefficient calculation unit 24 is an example of the prediction means and the correction means, and the roll gap set value calculation is performed. The unit 25 is an example of the setting means, and the control unit 26 is an example of the control means.

FIG. 2 is a flow chart showing a procedure example of the plate thickness control method according to the embodiment realized in the rolling system 100. The plate thickness control device 2 executes the information processing shown in the figure according to the program.

The processes S11 to S17 shown in the figure are executed in the setup control in the inter-pass time from the metal-off of the k-1 pass to the metal-in of the k-pass in multi-pass rolling. k is an arbitrary natural number and represents the number of passes. The k-pass is a target path for which the rolling load is predicted, and is, for example, the final path. The k-1 path is the path immediately before the target path.

The multi-pass rolling of the present embodiment is controlled rolling, and there is a target steel sheet temperature at the end of the final pass. Therefore, for example, the length of the inter-pass time from the end of the pass immediately before the final pass to the start of the final pass is adjusted in order to achieve the target steel sheet temperature. The processes of S11 to S17 may be executed only between the paths whose lengths are adjusted, or may be executed between the paths.

First, when the k-1 pass is completed, the plate thickness control device 2 calculates the load prediction value of the next k pass (S11: processing as the load prediction value calculation unit 21). Specifically, the plate thickness control device 2 calculates the _{load prediction value P pred based on the following equation (a).}

Further, it is assumed that the deformation resistance k _f is given by, for example, Shida's equation (the following equation (b)).

By the way, in the calculation of the load prediction value of S11, the softening phenomenon of the steel sheet is not taken into consideration. That is, the above formula (b) does not include softening influencing factors that affect the softening of the steel sheet, such as the time between passes. Therefore, when a softening phenomenon occurs between passes, such as when the time between passes is long, the rolling load may be overestimated.

Therefore, in the present embodiment, in S12 to S15 described below, the decrease in the rolling load due to the softening phenomenon of the steel sheet is corrected by using Gaussian process regression.

Next, the plate thickness control device 2 acquires the request point data (S12: processing as the request point data acquisition unit 22). The required point data is rolling-related data related to the target path and the like, and includes data of softening influencing factors that affect the softening of the steel sheet.

For example, the request point data includes one or more of the following data representing the rolling conditions of the k-pass which is the target pass.
・ Steel plate temperature predicted value ・ Sheet thickness command value ・ Reduction rate ・ Temperature difference

Here, the temperature difference is the temperature difference between the steel plate temperature at the end of the k-1 pass and the target steel plate temperature of the k-1 pass. This temperature difference corresponds to the inter-pass time from the end of the k-1 pass to the start of the k-pass, which is considered to have a particular effect on the softening of the steel sheet.

Further, the request point data may include one or more of the following data representing the rolling conditions of the k-1 pass, which is the pass immediately before the target pass.
・ Time between passes (end of k-2 pass to start of k-1 pass)
・ Actual steel sheet temperature value (on the surface of the steel sheet)
・ Calculated steel sheet temperature (at the center of the sheet thickness direction)
・ Plate thickness command value ・ Reduction rate

Further, the request point data may include one or more of the following data representing the rolling conditions of the k-2 pass, which is the pass two before the target pass.
・ Time between passes (end of k-3 pass to start of k-2 pass)
・ Actual steel sheet temperature value (on the surface of the steel sheet)
・ Calculated steel sheet temperature (at the center of the sheet thickness direction)
・ Plate thickness command value ・ Reduction rate

Next, the plate thickness control device 2 acquires stored data related to past rolling from the storage unit (S13: processing as the stored data acquisition unit 23). Specifically, the stored data includes a plurality of sets of training data and hyperparameters.

The training data includes an error index representing an error between the actual value and the predicted value of the rolling load in the past rolling as an objective variable, and rolling-related data of the same items as the above required point data as an explanatory variable.

FIG. 3 is a diagram showing an example of training data. N sets of training data include a one-dimensional objective variable and a D-dimensional explanatory variable. In the table, y _i represents the i-th component of y. x _{i (d)} represents the dth item of the explanatory variable of the i-th group training data.

Hyperparameters can be calculated in advance by type 2 maximum likelihood estimation or the like using training data.

Next, the plate thickness control device 2 generates a prediction model based on the stored data acquired in S13, and predicts an error index from the required point data acquired in S12 according to the prediction model (S14: softening correction). Processing as the coefficient calculation unit 24).

The predicted error index is the softening correction factor. The error index is the ratio of the actual value and the predicted value of the rolling load in the past rolling, and represents the decrease in the rolling load due to the softening phenomenon of the steel sheet. The error index is not limited to the ratio of the actual value and the predicted value, but may be a difference.

The prediction model is generated using the error index included in the training data as the objective variable and the rolling-related data of the same items as the required point data as the explanatory variables. As the prediction model, a non-linear regression model such as a Gaussian process regression model is suitable.

Next, the plate thickness control device 2 corrects the load prediction value calculated in S11 according to the error index predicted in S14 (S15: processing as the softening correction coefficient calculation unit 24). That is, the load prediction value calculated by the existing deformation resistance calculation formula that does not consider the softening phenomenon of the steel sheet is corrected by using the error index representing the decrease in the rolling load due to the softening phenomenon of the steel sheet.

The specific method for correcting the predicted load value is as follows. First, the ratio P / P _pred (hereinafter referred to as load ratio R, which corresponds to an error index) between the actual value P of the rolling load and the predicted load value P _{pred according to the above formula (a) is used as the objective variable, and the softening influencing factor is used.} Using the rolling-related data included as an explanatory variable, the load ratio is predicted from the required point data by the Gaussian process regression of the following equation (c). Next, by multiplying the load ratio predicted value R _pred by the load predicted value P _pred , the corrected rolling load P _corr is obtained by the following equation (d).

When the number of training data is large, the ^{amount of calculation for calculating K-1} in the above equation (c) becomes large, and the time required for predicting the load ratio increases, which may make it unsuitable for online execution. ^{Therefore, α = K -1} y may be generated in advance offline from the training data and the super parameters, and _{the load ratio predicted value R pred} may be calculated online using the following equation (e).

The above equation (d) for calculating the corrected rolling load P _corr can be modified as the following equation (f). Therefore, it can be interpreted that the above equation (d) reflects the decrease in deformation resistance due to the softening phenomenon by multiplying the deformation resistance by the correction coefficient COF _pred.

Next, the plate thickness control device 2 _{calculates the roll gap set value of the setup control based on the corrected rolling load P corr} (S16: processing as the roll gap set value calculation unit 25). The roll gap set value is calculated by, for example, the gauge meter formula of the following formula (g).

Next, the plate thickness control device 2 controls the roll gap of the rolling mill 1 based on the roll gap set value calculated in S16 (S17: processing as the control unit 26). After that, the metal-in of k-pass is carried out.

According to the embodiment described above, it is possible to improve the prediction accuracy of the rolling load by correcting the decrease in the rolling load due to the softening phenomenon of the steel sheet by using the Gaussian process regression.

Although the embodiments of the present invention have been described above, the present invention is not limited to the embodiments described above, and it goes without saying that various modifications can be made by those skilled in the art.

1 Rolling machine, 11, 12 work rolls, 2 Plate thickness control device, 21 Load prediction value calculation unit, 22 Request point data acquisition unit, 23 Storage data acquisition unit, 24 Softening correction coefficient calculation unit, 25 Roll gap setting value calculation unit , 26 Control unit, 100 Rolling system

Claims (13)

It is a plate thickness control device for rolling mills that perform rolling in multiple passes.
A calculation unit that calculates the predicted rolling load of the target path,
An acquisition unit that acquires rolling-related data related to the target pass or a pass before that, and
Prediction to predict the error index from the acquired rolling-related data according to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable and the rolling-related data as the explanatory variable. Department and
A correction unit that corrects the calculated rolling load prediction value according to the predicted error index, and a correction unit.
A setting unit that calculates the set value of the roll gap of the rolling mill based on the predicted value of the corrected rolling load, and a setting unit.
A plate thickness control device. The rolling-related data includes data on softening influencing factors that affect the softening of steel sheets.
The plate thickness control device according to claim 1. The rolling-related data includes data corresponding to the time between passes from the end of the pass immediately before the target pass to the start of the target pass.
The plate thickness control device according to claim 1 or 2. The data corresponding to the inter-pass time is the data of the temperature difference between the steel plate temperature at the end of the previous pass and the target steel plate temperature of the target pass.
The plate thickness control device according to claim 3. The rolling-related data includes one or more data selected from a steel plate temperature predicted value, a plate thickness command value, and a rolling reduction of the target pass.
The plate thickness control device according to any one of claims 1 to 4. The rolling-related data includes one or more data selected from the steel plate temperature of the previous pass, the plate thickness command value, the rolling ratio, and the time between passes with the two previous passes.
The plate thickness control device according to any one of claims 1 to 5. The rolling-related data includes one or more data selected from the steel plate temperature, the plate thickness command value, the rolling ratio, and the interpass time with the three previous passes of the target pass.
The plate thickness control device according to any one of claims 1 to 6. The prediction unit acquires the error index and rolling-related data in the past rolling from the storage unit, and generates the prediction model.
The plate thickness control device according to any one of claims 1 to 7. The prediction unit uses the prediction model generated in advance based on the error index in the past rolling and the rolling-related data.
The plate thickness control device according to any one of claims 1 to 7. An acquisition unit that acquires rolling-related data regarding the target pass or the previous pass of rolling of multiple passes by a rolling mill, and
Prediction to predict the error index from the acquired rolling-related data according to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable and the rolling-related data as the explanatory variable. Department and
A correction unit that corrects the predicted value of the rolling load of the target path according to the predicted error index.
A plate thickness control device. Calculate the predicted value of the rolling load of the target pass for rolling of multiple passes by the rolling mill.
Acquire rolling-related data regarding the target pass or the pass before that,
According to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable and the rolling-related data as the explanatory variable, the error index is predicted from the acquired rolling-related data.
The calculated predicted value of the rolling load is corrected according to the predicted error index.
The set value of the roll gap of the rolling mill is calculated based on the predicted value of the corrected rolling load.
Plate thickness control method. Calculating the predicted rolling load of the target pass for rolling of multiple passes,
Acquiring rolling-related data regarding the target pass or the pass before it,
Predict the error index from the acquired rolling-related data according to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable and the rolling-related data as the explanatory variable. ,
Correcting the calculated predicted value of rolling load according to the predicted error index,
To calculate the set value of the roll gap of the rolling mill based on the predicted value of the corrected rolling load.
A program that causes a computer to run. A rolling mill that rolls multiple passes and
A calculation unit that calculates the predicted rolling load of the target path,
An acquisition unit that acquires rolling-related data related to the target pass or a pass before that, and
Prediction to predict the error index from the acquired rolling-related data according to the prediction model generated with the error index representing the error between the actual value and the predicted value of the rolling load in the past rolling as the objective variable and the rolling-related data as the explanatory variable. Department and
A correction unit that corrects the calculated rolling load prediction value according to the predicted error index, and a correction unit.
A setting unit that calculates the set value of the roll gap of the rolling mill based on the predicted value of the corrected rolling load, and a setting unit.
A control unit that controls the roll gap of the rolling mill based on the calculated set value, and
Equipped with a rolling system.

Images